Pinyon and juniper tree encroachment is a growing problem in the western United States. As conifers expand they replace valuable sagebrush and grassland communities. The associated increases in fuel loads may lead to large, high severity fires and subsequent weed dominance. Ground methods used to measure tree cover and assess potential fire danger and ecosystem degradation are often time consuming and labor intensive. The extraction of tree cover from remotely-sensed imagery may provide a more rapid and economical method than ground measurements alone. Our objective was to evaluate three object-based image analysis (OBIA) software programs: Feature Extraction (ENVI Zoom 4.5), eCognition (Trimble), and Feature Analyst (Visual Learning Systems) capability to extract tree cover from NAIP imagery (1 m spatial resolution). We then evaluated the relationship between our remotely sensed tree cover estimates from each of the software programs with two ground measurements methods (tree-crown diameter and line-point intercept). Data was collected on 300, 30 by 33 meter subplots throughout Utah. Our initial results for our OBIA tree cover estimates were on average 5% greater than the tree-crown diameter measurements and 15% greater than our line-point intercept measurements. These differences could be due to field measurement errors, variance in NAIP image quality, shifts in subplots due to GPS inaccuracies, and/or that our object-based image analysis techniques over-segments pixels and may classify small portions of bare ground, shrubs, or grasses as trees. Analysis is ongoing for biomass estimation and should be completed shortly.
Oral presentation and poster titles, abstracts, and authors from the Society for Range Management (SRM) Annual Meetings and Tradeshows, from 2013 forward.